Little Golem TwixtPP Historical Data on Wins by Opening Move

  b—wc—vd—ue—tf—sg—rh—qi—pj—ok—nl—m
1—24(17)
[ 0.0]
57
(7)
[ 0.0]
43
(8)
[ 0.0]
24
(12)
[ 0.0]
52
(10)
[ 0.0]
18
(10)
[ 0.0]
14
(12)
[ 0.0]
25
(14)
[ 0.0]
87
(17)
[ 0.0]
59
(27)
[ 0.0]
40
(32)
[ 0.0]
40
2—23(26)
[ 0.0]
45
(18)
[ 5.6]
24
(29)
[ 3.4]
27
(32)
[ 9.4]
57
(18)
[11.1]
51
(9)
[11.1]
63
(8)
[ 0.0]
15
(6)
[16.7]
40
(7)
[ 0.0]
25
(8)
[ 0.0]
47
(42)
[ 2.4]
59
3—22(28)
[ 7.1]
59
(727)
[20.6]
48
(7709)
[30.0]
50
(3949)
[27.0]
47
(1772)
[24.1]
46
(252)
[27.0]
45
(80)
[28.7]
36
(52)
[26.9]
61
(33)
[30.3]
51
(53)
[28.3]
44
(157)
[22.3]
48
4—21(65)
[ 7.7]
45
(1891)
[33.2]
55
(2477)
[41.1]
55
(2257)
[45.7]
52
(1270)
[51.1]
54
(1043)
[52.3]
45
(372)
[40.6]
53
(244)
[39.3]
44
(98)
[36.7]
47
(64)
[34.4]
68
(109)
[24.8]
46
5—20(35)
[11.4]
34
(1048)
[29.0]
55
(567)
[44.4]
60
(1168)
[50.2]
57
(350)
[51.4]
56
(259)
[52.9]
50
(426)
[53.5]
53
(162)
[50.0]
46
(186)
[52.2]
46
(142)
[37.3]
46
(213)
[40.4]
43
6—19(40)
[15.0]
48
(479)
[29.2]
58
(620)
[47.3]
59
(175)
[52.0]
64
(473)
[55.8]
54
(209)
[52.6]
52
(118)
[57.6]
56
(108)
[50.0]
60
(121)
[47.9]
53
(90)
[44.4]
52
(162)
[50.0]
53
7—18(12)
[ 8.3]
36
(209)
[24.4]
60
(260)
[48.5]
66
(127)
[51.2]
51
(132)
[56.8]
67
(241)
[58.9]
57
(124)
[57.3]
58
(79)
[51.9]
66
(98)
[53.1]
56
(95)
[53.7]
64
(192)
[40.6]
54
8—17(30)
[13.3]
36
(456)
[27.4]
59
(163)
[47.2]
52
(146)
[50.7]
71
(81)
[46.9]
61
(89)
[53.9]
63
(176)
[52.3]
57
(108)
[50.9]
62
(61)
[47.5]
49
(71)
[54.9]
59
(113)
[46.0]
55
9—16(15)
[ 6.7]
52
(60)
[23.3]
58
(72)
[51.4]
58
(43)
[46.5]
56
(44)
[45.5]
63
(76)
[57.9]
58
(85)
[51.8]
63
(890)
[55.5]
62
(149)
[53.7]
61
(91)
[48.4]
62
(102)
[50.0]
52
10—15(8)
[12.5]
26
(56)
[16.1]
38
(57)
[47.4]
64
(51)
[45.1]
61
(99)
[52.5]
54
(58)
[55.2]
72
(74)
[63.5]
48
(126)
[48.4]
60
(222)
[61.3]
63
(128)
[54.7]
57
(148)
[50.7]
57
11—14(4)
[ 0.0]
32
(84)
[26.2]
42
(46)
[43.5]
48
(39)
[46.2]
61
(38)
[39.5]
27
(68)
[42.6]
49
(88)
[45.5]
48
(106)
[57.5]
55
(134)
[61.2]
55
(223)
[61.9]
63
(341)
[56.9]
59
12—13(78)
[19.2]
49
(474)
[24.3]
51
(162)
[53.1]
67
(62)
[46.8]
56
(72)
[44.4]
71
(135)
[45.9]
61
(110)
[56.4]
64
(121)
[56.2]
62
(174)
[61.5]
59
(399)
[59.4]
59
(1390)
[64.8]
61
For 42287 games completed by 2013-03-24.

Data excludes games where either player has a Ratings Deviation (RD) greater than 180 (so that the rating can be thought to be reasonably accurate) and games that were forfeits, draws, finished with fewer than 10 pegs or more than 100 pegs, involved one of the 25 worst players overall, and games where the player played against himself.

Since an empty Twixt board is symmetrical, the table shows only one quadrant, but includes data reflected from all four. The axes show coordinates from all quadrants—for example, c3 is equivalent to v3, c22, and v22. The "a—x" column is not shown because it is illegal to place your peg behind the enemy border. The "1—24" row is behind the first player’s border.

Numbers in parentheses on the top of each square are the total number of times that a player played that position as their first move.

Numbers in brackets in the middle of each square are the percentage of times that the second player swapped the move. If it was swapped, the second player’s outcome is counted on the square.

Scores are calculated using a method designed to reduce the influence of skills differences between players. An high number is added or subtracted for an upset, or a low number for the expected outcome. That number is the "Win Expectancy," as calculated by the Glicko rating method. For example: if Player A has a .8 win expectancy over Player B, then in 5 games, A is expected to win 4 games, and B should win 1. If player A led with f3 in each of those 5 games, then the f3 spot would gain .2 points for each victory, and lose .8 points for the loss. The net score over 5 games would be 0, indicating that f3 is a fairly neutral spot. The total is then divided by the sum of the absolute values of these scores, which yields a score from -1 to 1, which is then normalized to a scale from 0 to 100. The end result is an estimation of the strength of the spot, where 50 is exactly neutral.

Color coding is as follows:
Red backgrounds indicate the strongest first moves (above 60).
Orange backgrounds are on numbers around 55 (about as strong as playing White in Chess).
Yellow backgrounds are on the most neutral spots.
Green backgrounds are on numbers around 45.
Blue backgrounds indicate the weakest first moves (below 40).
Black backgrounds indicate spots where there is not enough data; for each spot with fewer than 200 games started on the spot, the background is darkened linearly.

A simple reading of this table is that if you are the first player, you should play where the number is closest to 50 with the best data (bright yellow), and if you are the second player, you should swap if the number is greater than 50. More realistically, if you are the first player, you may want to consider the most advantageous (reddest) position that you think your opponent will not swap, or the weakest (bluest) position that you think your opponent will swap, and if you are the second player, you might consider whether you know a good response to the first move, or whether you would rather leave it up to the first player. Keep in mind also that if a spot is orange, that does not mean that "swap" was historically the best response—it means that "swap" was historically better than the average response.

Disclaimer: Past performance is not a predictor of future results. These stats are influenced by the skills and preferred opening moves of the Twixt players that have played on Little Golem.

Have fun!